Seismic structure of an oceanic core complex at the Mid-Atlantic Ridge, 22°19’N Anke Dannowski, Ingo Grevemeyer, C.R Ranero, Georges Ceuleneer, Marcia Maia, Jason Morgan Phipps, Pascal Gente

To cite this version:

Anke Dannowski, Ingo Grevemeyer, C.R Ranero, Georges Ceuleneer, Marcia Maia, et al.. Seismic structure of an oceanic core complex at the Mid-Atlantic Ridge, 22°19’N. Journal of Geophysical Research : Solid Earth, American Geophysical Union, 2010, 115, pp.B07106. ￿10.1029/2009JB006943￿. ￿insu-00563250￿

HAL Id: insu-00563250 https://hal-insu.archives-ouvertes.fr/insu-00563250 Submitted on 4 Feb 2011

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, B07106, doi:10.1029/2009JB006943, 2010 Click Here for Full Article

Seismic structure of an oceanic core complex at the Mid‐Atlantic Ridge, 22°19′N Anke Dannowski,1 Ingo Grevemeyer,1 Cesar R. Ranero,2 Georges Ceuleneer,3 Marcia Maia,4 Jason Phipps Morgan,5 and Pascal Gente4 Received 1 September 2009; revised 8 January 2010; accepted 25 January 2010; published 24 July 2010.

[1] We present results from a seismic refraction and wide‐angle experiment surveying an oceanic core complex on the Mid‐Atlantic Ridge at 22°19′N. Oceanic core complexes are settings where petrological sampling found exposed lower crustal and upper mantle rocks, exhumed by asymmetric crustal accretion involving detachment faulting at magmatically starved ridge sections. Tomographic inversion of our seismic data yielded lateral variations of P wave velocity within the upper 3 to 4 km of the lithosphere across the median valley. A joint modeling procedure of seismic P wave travel times and marine gravity field data was used to constrain crustal thickness variations and the structure of the uppermost mantle. A gradual increase of seismic velocities from the median valley to the east is connected to aging of the oceanic crust, while a rapid change of seismic velocities at the western ridge flank indicates profound differences in lithology between conjugated ridge flanks, caused by un‐roofing lower crust rocks. Under the core complex crust is approximately 40% thinner than in the median valley and under the conjugated eastern flank. Clear PmP reflections turning under the western ridge flank suggest the creation of a Moho boundary and hence continuous magmatic accretion during core complex formation. Citation: Dannowski, A., I. Grevemeyer, C. R. Ranero, G. Ceuleneer, M. Maia, J. P. Morgan, and P. Gente (2010), Seismic structure of an oceanic core complex at the Mid‐Atlantic Ridge, 22°19′N, J. Geophys. Res., 115, B07106, doi:10.1029/2009JB006943.

1. Introduction regime compared to magmatic segments [Smith et al., 2006; ‐ ‐ Escartin et al., 2008] and occurs generally at ridge crest [2] Lithospheric accretion along the slow spreading Mid discontinuities. However, within the last 5 years OCCs have Atlantic Ridge (MAR) is driven by two contrasting modes: been recognized to occur nearly randomly along some (1) asymmetric accretion involving active detachment spreading segments [Smith et al., 2006, 2008]. faulting and core complex formation and (2) symmetric [3] Spatial and temporal variation in magmatic and tec- magmatic accretion with abyssal hill formation on both tonic processes leads to a high structural variability at OCCs conjugated ridge flanks [Escartin et al., 2008]. Oceanic core [Tucholke and Lin, 1994; Cann et al., 1997; Escartin et al., complexes (OCCs) are lower crustal or upper mantle rocks 2003; Smith et al., 2006], resulting in different models exhumed at the seafloor by long‐lived detachment faults. At explaining the observations. Tucholke et al. [1998] suggest some settings they may accommodate extension for 1–2m.y. that core complexes form by a detachment fault rooting [Tucholke and Lin, 1994; Tucholke et al., 1998]. Character- ‐ at the base of the lithosphere. The onset of amagmatic istic features of OCCs are blocky topography with dome extension results in detachment faulting and the formation of shaped structures and often spreading‐parallel corrugated a detachment that accretes lithospheric mantle and exposes surfaces [Cann et al., 1997; Blackman et al., 1998; Tucholke the deep lithosphere. In an alternative model Dick et al. et al., 1998]. Active detachment faulting is associated with [2000] proposed a detachment fault rooting at or near a near constant levels of seismic activity in a colder thermal melt‐rich zone near the spreading axis. Continuous mag- matic accretion occurs during detachment faulting, expos- ing mainly gabbros that are deformed at high temperatures 1Leibniz Institute of Marine Science at University of Kiel (IFM‐ and in the presence of melt. A third conceptual model GEOMAR), Kiel, Germany. [Escartin et al., 2003] localizes the detachment fault rooting 2ICREA at Institut de Ciencias del Mar, CSIC, Barcelona, Spain. in the shallow lithosphere. During extension active gabbro 3 ‐ Toulouse University, Observatoire Midi Pyrénées, Toulouse, France. intrusions and dyke injections occur episodically across the 4Université Européenne de Bretagne, Brest, France. 5EAS Department, Cornell University, Ithaca, New York, USA. fault zone. They will be exposed during continued exten- sion. The alteration front may correspond to a rheological Copyright 2010 by the American Geophysical Union. boundary. 0148‐0227/10/2009JB006943

B07106 1of15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

[4] These models would predict very different crustal and [7] Figure 1b shows the 22°19′N core complex. Its eastern upper mantle structure at the OCC and the hanging wall flank dips at an angle of 42°. The fault intersects the valley preserved in the median valley and at the conjugated ridge surface roughly 6–7 km away from the ridge axis. The next flank. The main difference between the models would be the fault block to the east is suggested to be the active fault geometry of the crust/mantle boundary (i.e., seismic Moho). bounding the axial valley. The break‐away is difficult to While in the amagmatic extension model [Tucholke et al., define using the existing bathymetric data. However, the 1998], the Moho would reach the seafloor under the core western end of the dome occurs 20–25 km off the ridge axis complex; the Moho would occur several kilometers below and may suggest that the fault system is active for about the seafloor in the melt‐assisted extension model [Dick et al., 2 m.y. Several ridge‐parallel features crosscut the core 2000]. For the episodically magmatic extension with an complex and may represent rotated fault (rafted) blocks alteration front [Escartin et al., 2003], a seismic velocity covering the surface of the detachment fault or secondary gradient would be expected rather than a sharp contrast in ridge parallel fault zones. Further west abyssal hills are velocity. aligned to the core complex, though westward a domal [5] A number of seismic surveys studied OCCs at active structure, possibly a fossil core complex occurs. spreading ridges like the Atlantis Bank at the Southwest Indian Ridge (SWIR) [Muller et al., 2000; Dick et al., 2000], the Atlantis Massif at the MAR near 30°N [Canales et al., 3. Geology of the Core Complex 2004] (MAR), the TAG area at the MAR near 26°N, the [8] The geological structure of the Mid‐Atlantic Ridge at Mid‐Atlantic Ridge at Kane (MARK) area at the MAR near 22°19′N could be derived from two dredge campaigns 23°20′N[Canales and Collins, 2000; Canales and Collins [Karson et al., 1987; Cannat et al., 1995] and previously et al., 2008; de Martin et al., 2007], and at the MAR 5°S unpublished data from a dive of the French submersible [Planert et al., 2010]. However, they concentrate on the Nautile. Figure 2 presents results of these studies. OCC itself and resolve only the upper 1–2 km. Here we will [9] During the TAMMAR cruise with R/V Nadir in 1996, use seismic data to study the crustal and upper mantle the submersible Nautile explored during its dive 21 the structure surveying the seismic velocity structure of an OCC southeastern slope of the 22°19′N dome along a SE‐NW and its associated hanging wall on the conjugated rift trending ∼4 km long transect (Figure 2). Within the median shoulder, crossing the Mid‐Atlantic Ridge at 22°19′N. A valley the seabed is heavily sedimented. Scattered and iso- joint modeling approach using both seismic and gravimetric lated angular blocks of serpentinite fallen down from the data resolves both the crustal and upper mantle structure. nearby cliffs occurred rarely. Close to the foothill, a narrow ∼E‐W trending ridge of mylonitic and amphibolitized gabbros emerged from the sediments. At a depth of 2960 m 2. Tectonic Setting serpentinized peridotites crop out. Serpentinites are intensely fractured. Further up slope, at 2230 m, a brecciated zone is ‐ [6] At the Mid Atlantic Ridge to the south of the Kane detected. The breccia is made of fragments of serpentinized transform fault boundary the North American and African peridotites and of metadiabase embedded in a fine matrix plates separate at a rate of 2.5 cm/yr [Gente et al., 1995]. made of serpentine, chlorite, and likely clay minerals. The The MARK area to the south of the Kane transform fault has thickness of the brecciated zone reaches several hundred been targeted by a number of seismic refraction studies, meters and its general orientation is N‐S with a subvertical yielding a complex and heterogeneous lithospheric archi- dip. This “cold” contact separates the serpentinites from the tecture dominated by amagmatic extension at the OCC to overlying outcrops of diabase dykes and basalts. The ori- the south of the Kane transform fault [Canales et al., 2008; entation of the dykes is N‐S with a high westward dip (i.e., Dick et al., 2008; Toomey et al., 1988] and a magmatically parallel to the one of the diaclases cutting the serpentinites). dominated mode of lithospheric accretion away from the At a depth of 2050 m a subhorizontal E‐W trending shear transform fault [Purdy and Detrick, 1986; Canales and zone crosscuts the diabase dykes that are pervasively Collins, 2000]. Further south an area of high seismic transformed into an assemblage of chlorite and other green ‐ activity characterizes the Mid Atlantic Ridge between 22° schist facies minerals. The sampled gabbros and serpenti- ′ ′ 10 N and 22°40 N[Escartin et al., 2008]. This area is sit- nites are described in detail in the Appendix. Ghose et al. uated north of the highly magmatic southward propagating [1996] gave petrological data on the ultramafic rocks ′ TAMMAR segment. Centered roughly at 22°17 N, a short dredged south of this core complex at the 22°10′N discon- ridge segment [Gente et al., 1995] that existed for roughly tinuity that terminates the segment end toward the south. 5 m.y. occurs. Segment boundaries to the north and south, [10] In contrast to the ultramafic and gabbroic rocks that however, are not well defined. The segment length can best have been sampled on the western high, sampling of the be defined by an anomalously deep seafloor, which is eastern conjugated ridge flank and in the median valley – roughly 500 1000 m deeper that the seafloor to the north suggest that the lithology is dominated by basalts and dia- ‐ and south. A dome shaped structure on the western ridge bases [Karson et al., 1987; Cannat et al., 1995], supporting ′ flank at 22°19 N is interpreted as core complex [Cannat et al., a mode of asymmetric lithospheric accretion. Thus, core ‐ 1995] caused by detachment faulting with spreading parallel complex formation unroofed the footwall of lower crustal corrugations (Figure 1). The core complex is located and upper mantle rocks on the western flank and transferred ‐ approximately 10 km off axis to the west,and its summit the hanging wall of upper crust to the east. Seafloor occurs at the magnetic anomaly 2, indicating an age of about topography supports this interpretation. Blocky topography 1.8 m.y. [Gente et al., 1995]. characterizes the western ridge flank, while the eastern

2of15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Figure 1. Bathymetric maps of the study area. (a) Overview of the global tectonic features. The red rect- angle marks the survey field south of MARK area. (b) Domed structure in a 3‐D view. This is the key feature of the study and is interpreted as an oceanic core complex. (c) Overview of the seismic line with its deployed instruments (15 Ocean Bottom Hydrophones). conjugated flank shows linear abyssal hills and small basins ship was navigated with GPS. Before deploying the OBHs, (Figure 1c). all clocks were rated against a GPS DCF77 time signal. After recovery, the OBH internal clock was rated against GPS time to determine the drift, in order to calculate time 4. Seismic Experiment and Gravity Data corrections. Timing errors were generally less than 30 ms [11] START HERE In December of 2003 and January of and were assumed to be related to a linear drift over the time 2004 the COSTMAR cruise (M60/2) was conducted aboard of deployment. Data passed an antialiasing filter and were the German research vessel Meteor at the Mid‐Atlantic continuously recorded with a sampling rate of 100 or 125 Hz Ridge between 21°30′N and 22°30′N. Profile p08 crossed on all OBHs. The data were played back and split into single the 22°19′N core complex. Fifteen ocean bottom hydro- shot records stored as receiver gathers in SEG‐Y format. phones [Flueh and Bialas, 1996] were deployed along the Spectral analysis and filter tests show that the seismic 150 km long seismic transect, sampling the structure of a energy is in a band ranging from 4 to 20 Hz. We ran this test portion of the Mid‐Atlantic Ridge. Two 32 L Bolt air guns, for both near‐offset and far‐offset traces and chose a time‐ operating at a pressure of 130 bar, were used as seismic and offset‐dependent filtering approach. In addition, a pre- sources. Guns were fired at a shot interval of 60 s, resulting dictive deconvolution was applied and previously amplitudes at a ship’s speed of about 4 knots over ground into a shot were multiplied by distance to partly compensate the spacing of approximately 120 m. Ocean bottom hydro- spherical divergence, simultaneously showing the level of phones (OBHs) were deployed by free fall, using Global both seismic signal and ambient noise. Data examples can Positioning System (GPS) for drop point positioning. The be found in Figure 3. instrument locations were further constraints using water [12] Gravity data used in our study were obtained in 1991 wave arrival times from air gun shots collected while the during the SEADMA1 cruise with the French research

3of15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Figure 2. Bathymetric map of the core complex showing the location of the Tammar‐Dive21 and the seismic line including stations that cover the massif. Dredges from studies of the MARK area are included. vessel l’Atalante [Gente et al., 1991, 1995] with a KSS30 tomography of crustal phases (Pg) to yield the seismic Bodenseewerk gravity meter [Maia and Gente, 1998]. structure of the crust. In a second step, however, we included Moho reflections; constraining crustal thickness were PmP phases that have been sampled. Constraints on 5. Modeling Procedure the crustal thickness along the entire profile were further [13] Seismic data recorded along the profile have gener- obtained using forward modeling of upper mantle turning ally been of excellent quality. However, it is interesting to rays (Pn) and gravity data. The seismic and gravimeric note that seismic wide‐angle reflection data have just been Moho were derived using an iterative approach, updating recorded on a few stations, though refracted arrivals of crustal thickness simultaneously for both data sets. In addi- energy turning in the uppermost mantle have been imaged to tion, the modeling procedure yielded upper mantle seismic offsets of 60–80 km. Thus, record sections generally lack velocities and densities. high‐quality wide‐angle reflection phases from the crust/ mantle boundary (PmP), perhaps related to a weak trace‐to‐ 5.1. Results From Tomographic Inversion trace correlation of seismic phases due to scattering and [14] Seismic data at small offsets of up to 30 km have a diffractions caused by the rough seafloor relief. However, high signal‐to‐noise ratio (Figure 3), allowing a clear lateral changes in the lithology at Moho depth may also identification of a crustal refraction branch (Pg). The 2305 govern the occurrence of reflected arrivals. The fact that hand‐picked travel times (picking error < 40 ms) show only a few record sections provided clear PmP reflections as a strong scattering at equal shot‐receiver distances. This secondary arrivals in a credible manner had governed the strong variation in the travel curves is perhaps related to inversion and modeling procedure. Thus, seismic travel time both the rough seafloor topography and a strong variation in tomography without wide‐angle constraints on the geometry the shallow seismic structure along the profile, as reported of the seismic Moho may suffer from a trade‐off between elsewhere for OCCs [Canales et al., 2008]. changes in seismic velocity and layer geometry. Therefore, [15] We used the tomographic method of Korenaga et al. due to the complexity of the area, different modeling tech- [2000] to invert P wave travel times, yielding a seismic niques like seismic tomographic inversion, ray trace mod- velocity model for the uppermost 3–5 km of the crust. A eling, and gravity modeling were used to resolve the crustal regular grid with uniform horizontal grid cells of 200 m and and upper mantle structure. We used seismic travel time a vertical increasing spacing from 50 to 170 m was used.

4of15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Figure 3. Seismic sections of reduced velocity to 8 km/s are shown as data examples of the seismic line P08. The signal‐to‐noise ratio is high up to an offset of 40 km, and the refracted waves of the crust and the mantle are highly visible at most of the stations and up to 80 km for some OBHs. However, the mantle reflection phase cannot be continuously identified over all stations. Interpreted seismic arrivals have been labeled: Pg (turning rays within the crust), Pn (turning rays in the upper mantle), and PmP (reflected rays at the Moho).

The grid hangs below the seafloor and was parameterized at [16] A strong lateral variation in crustal structure is well a spacing of 500–600 m. During inversion we used layer displayed in the seismic velocity model (Figure 4). The stripping, inverting first for the upper crustal arrivals or near eastern flank of the ridge shows typical features for normal offset picks, and proceeded with the larger offset picks, oceanic crust with a high velocity gradient in layer 2 and a leading to a robust model of the refracted data for the crust. relative constant thickness of 1.5–2.2 km. The estimated A standard 1‐D starting velocity model had been defined as velocity rises from 3.2 km/s at the top of igneous crust to reference model for 2‐D tomographic inversion. The 1‐D about 6.5 km/s at 2 km below the seafloor. The lower crust model was derived using constraints from previous seismic has a lower gradient, and velocities increase slowly to 7.0– refraction surveys at the MAR to the south of Kane trans- 7.2 km/s near the expected bottom of crust. In the median form boundary [Purdy and Detrick, 1986; Canales et al., valley, seismic velocities are slowest; low velocities of 2.3– 2000]. However, different starting models were tested to 2.4 km/s were observed in the first few hundred meters yield the robustness of inversion result. Tests included below the seafloor in the eastern part of the axial valley. several half‐space models with different velocity gradients Velocity reaches 6.5 km/s at the depth of 4 km below the and several layered models. The final solution, however, seafloor with a low velocity gradient in the lower crust. depended only weakly on the starting model. Further west, approaching the dome‐shaped core complex,

5of15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Figure 4. (a) Preferred final P wave velocity model of the upper crust after five iterations. The interval of the contour lines is 0.5 km/s. Crustal picks up to an offset of 20 km and wide‐angle reflection arrivals have been used for the inversion tomography. The upper part of this figure, the bathymetric map, shows where stations and profile are situated. (b) Resolution of the model, the ray coverage of the tomographic inversion. (c) Velocity perturbations from average oceanic crust at the MAR.

6of15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Figure 5. Travel time fits for the three data examples in Figure 3. Observed (vertical bars) and predicted (colored dots) travel times for the final velocity model: Pg (red) and PmP (dark blue) from the inversion (red) and Pn from the forward modeling (light blue). Pn arrivals are constraining the upper mantle structure beneath the massif. At larger offsets, rays do penetrate, but the data quality is too low to gain trustable picks. velocities increase to values of 3–4 km/s at the seabed, and from an initial c2 = 13.08 after five iterations. The ray the uppermost crust is characterized by a steep vertical coverage (Figure 4b) constrains the model. velocity gradient. Lower crustal velocities of more than 7 km/s were imaged at already 1.5 km below the seafloor. 5.2. Combined Seismic and Gravity Modeling To the west of the core complex, crust is more similar to the [19] A benefit of using a ray tracing forward method was eastern flank, with a high velocity gradient in layer 2 and the application of a joint modeling approach using observed moderately increasing velocity in the lower crust, layer 3, travel times, including mantle arrivals (Pn), and the marine reaching velocities of 7.1–7.2 km/s at a depth of 4 km below free‐air gravity field. We used MacRay [Luetgert, 1992] to seafloor. generate a model satisfying both gravity and seismic data, [17] Crustal thickness could be derived at to two patches where Pn arrivals and gravity data provide useful constraints along the profile, where a number of Moho wide‐angle on the crustal thickness and structure of the uppermost reflections were observed. Generally, wide‐angle phases had mantle. good quality where the seafloor was smooth. Fortunately, [20] Seismic velocities have been converted to density OBH80 on the eastern flank provided a clear PmP phase, model using different empirical relationships for upper and indicating a crustal thickness of 6 km for the eastern flank. lower crust: r = 3.81–6/vp [Carlson and Herrick, 1990] for In addition, OBH66 to OBH71 on top of the core complex the upper crust; r = 0.375 + 0.375*vp [Birch, 1961] for provided PmP phases. Here, crustal thickness is ∼3.5 km, gabbroic crust found in the lower crust. The mantle was indicating approximately 40% thinner crust. initially assumed to have a constant density of 3.3 g/cm3. [18] In Figures 5 and 6 the observed and calculated travel For upper crust seismic velocities less than 4 km/s a value of times are compared. The final RMS misfit is 49 ms and the 2.4 g/cm3 has been assumed, using constraints from an c2 ∼ 2.01 for the crustal phases. The inversions converged ocean bottom gravity study at the Juan de Fuca Ridge

7of15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Figure 6. Travel time fits for all stations along profile 8. Observed (vertical bars) and predicted (colored dots) travel times for the final velocity model: Pg (red) and PmP (dark blue) from the inversion (red) and Pn from the forward modeling (light blue). Pn arrivals are constraining the upper mantle structure beneath the massif. At larger offsets, rays do penetrate, but the data quality is too low to gain trustable picks.

8of15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Figure 7. Combined seismic and gravity forward modeling: (a) gravity fit and (b) preferred final seismic velocity and density model. The colored section is the result of the joint near‐offset first arrival and wide‐ angle reflection tomographic inversion. The deeper structures as well as the density values result from forward modeling using the raytracing method [Luetgert, 1992]. The gray shaded area in the mantle shows the area with lowered densities. The red thickened lines of the crust–mantle boundary indicate the parts where the Moho is constraint by seismic wide‐angle reflection arrivals.

[Stevenson et al., 1994]. Figure 7 shows that there is not tracing and gravity modeling. It is interesting to note that a only a strong lateral variation in the density and velocity few PmP branches sampled both the normal roughly 6 km field at crustal levels but also density variations in the thick crust away from the core complex and the portion mantle. As a result of 934 picks of Pn arrivals distributed of thinner crust under the core complex, supporting crustal along the entire profile, we could image both the geometry thickness variations from the forward modeling. In addition, of the seismic Moho and the velocity structure of the the occurrence of PmP arrivals suggests that a well‐defined uppermost mantle. The modeling strategy was to start with a Moho discontinuity has been developed, indicating ongoing flat Moho. However, profound changes had to be introduced magmatic accretion during detachment faulting and hence to fit both Pn arrivals and gravity field data. Away from the supporting the model of Dick et al. [2000]. ridge crest, velocities of 7.8 km/s fit the Pn branches best. Below the median valley and the western ridge flank 6. Discussion velocities decrease to 7.5 km/s. [21] Potential field methods are nonunique, and the 6.1. Upper Crustal Structure ‐ interpretation of seismic results may suffer from the trade [23] The seismic data reported here is a rare data set off between seismic velocity and the thickness of a layer. covering both conjugated ridge flanks and hence the foot- Nevertheless, density contrast both at the water/seabed wall and hanging wall of a detachment fault. We obtained a interface and the Moho caused strong lateral variations of strong variation in seismic velocities within the upper 3– about 100 mGal in the observed gravity field. The topo- 4 km across the ridge axis. Velocities near the seabed vary graphic effect alone could not explain the observed anomaly by more than ±1.1 km/s. A distinct feature is a very rapid over the massif, supporting a shallower mantle (higher increase in P wave velocity where the median valley density material) underlying the core complex. The final approaches the western ridge flank. Velocities increase from model (Figure 7) shows reduced velocities and density in the about 2.3 to 4.5 km/s over a distance of 5–7 km and are mantle below the median valley and the core complex and fastest below the dome‐shaped core complex. On the con- reduced crustal thickness under the detachment fault, indi- jugated eastern ridge flank, however, seismic velocities cating crustal thinning caused by faulting. The RMS residual increase more gradually (Figure 4). for the gravity modeling is 5.22 mGal. The observed and [24] Seismic velocities may increase while crust ages. calculated travel times are plotted and compared in Figures 5 This phenomenon is related to hydrothermal precipitation of and 6. secondary alteration products into open pore spaces and [22] A few clear PmP branches have been observed and cracks, decreasing the porosity and hence increasing seismic were used to constrain the final model derived from ray velocities [e.g., Grevemeyer and Weigel, 1996; Carlson,

9of15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Figure 8. Velocity versus depth profiles. (left) Representation of the ridge axis (dashed line), the core complex to the west (solid), and the opposite side at same offset (dotted). The gray zone denotes the range of typical velocity profiles for 0–7 Ma oceanic crust after White et al. [1992]. This diagram supports the interpretation of a change in lithology. (right) Comparison to other studied OCC’s including sonic logs of the IODP hole U1309D and ODP hole 735B (IODP Lamont logging database), seismic models of the TAG area [de Martin et al., 2007], Atlantis Massif [Canales et al., 2008], Atlantis Bank [Muller et al., 2000], and MAR 5°S [Planert et al., 2010]. The sonic log data have been plotted twice; the brown plot has been moved downward to better compare the gabbroic section of our model with the gabbroic velocities from the drilling.

1998; Grevemeyer et al., 1999]. The impact of hydrothermal velocities of 4.5 km/s [Carlson and Miller, 2003]. Thus, activity on the seismic structure causes a gradual change in serpentinization may turn velocities of mantle peridiotites seismic velocities [Grevemeyer et al., 1999; Nedimović et al., into velocities typical for lower crustal or upper crustal rocks. 2008]. A rapid change as observed at the OCC, however, may Fracturing may even further decrease seismic velocities. support a change in lithology. Sampling supports this inter- Gabbroic rocks generally have velocities of 6.7–7.1 km/s. pretation. Dredges from the median valley floor and the Like for serpentinites, fracturing may also decrease values eastern ridge shoulder provided basaltic seafloor, while significantly. Thus, based on seismic velocity data alone, it sampling provided a wide range of rock types, including is difficult to discriminate between different rock types. serpentinites and gabbroic rocks from the core complex Nevertheless, Canales et al. [2008] found a correlation (Figure 2). between seismic velocities and lithology. They found that [25] Different models of core complex formation suggest areas where serpentinites have been dredged were charac- fundamental differences in lithology. Thus, while melt‐ terized by P wave velocities of <3.5 km/s, whereas veloci- assisted extension supports that the internal structure of the ties of >4 km/s generally occur where gabbros have been OCC is dominantly grabboic [Dick et al., 2000], amagmatic found. Our data at the 22°19′N core complex support fast extension may unroof mantle rocks [Tucholke et al., 1998]. velocities close to the seabed (3–4 km/s), reaching ∼7 km/s Petrologic interpretation of seismic velocities might be non- at 1.5 km below the seafloor (Figures 4 and 8). In the unique. For mantle rocks, for example, seismic P wave context of the interpretation by Canales et al. [2008], our velocity decreases with increasing degree of serpentinization. seismic data suggest that lower crustal rocks have been Velocities of >8 km/s characterize unaltered peridiotites, brought to shallow depth; thus, the core complex is suggested whereas 100% serpentinized peridotite shows decreased to be dominantly composed of gabbroic rocks (Figure 9). A

10 of 15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Figure 9. The geological interpretation of the seismic modeling. After initiation of the detachment the crust becomes approximately 40% thinner compared to normal oceanic crust. The mantle wells up while the detachment fault grows. To the east normal faulting occurs. The plate boundary moves from the center westward to the active fault of the core complex. The model is plotted with 2.5× vertical exaggeration. rather low seismic velocity gradient below the upper 1.5 km ultramafics [Cannat et al., 1995; Cannat, 1996], dominating supports these observations. An alteration front separating the lithology at segment ends. hydrated from dry mantle, as for example predicted by [28] Drilling into the Atlantis Bank at the Southwest Escartin et al. [2003], would rather cause a steep gradient Indian Ridge, ODP hole 735B, also provided mainly gabbroic instead of a first‐order discontinuity. Thus, the observation rocks from the core complex [Dick et al., 2000]. Dick et al. of wide‐angle reflections (our PmP arrivals) favors our [2000] proposed a detachment fault rooting at or near a interpretation of a lithological boundary separating crustal melt‐rich zone near the ridge axis, where a continuous from mantle rocks. magmatic crust is formed during detachment faulting, which exposes mainly gabbros. In such a scenario a Moho boundary 6.2. Crustal and Upper Mantle Structure is formed and therefore constrains our interpretation that lower crustal material forms the oceanic core complex, while [26] Recent drilling campaigns at OCCs [Blackman et al., 2006; Escartin et al., 2003; Ildefonse et al., 2007] obtained upper mantle has not been exposed. In other scenarios of core complex formation, where extension occurs during amag- mostly gabbroic rocks and hence support our interpretation. matic extension [Tucholcke et al., 1998] or where the For example, IODP hole U1309D penetrated roughly to a depth of >1.4 km and sampled dominantly gabbroic rocks of detachment fault is rooted in the shallow lithosphere [Escartin et al., 2003], we would not expect the formation of a highly primitive nature at the central dome of the Atlantis a profound density or velocity contrast at Moho depth. Thus, Massif to the north of the Atlantis transform fault boundary hydrothermal circulation in exposed, and a fractured mantle of the Mid‐Atlantic Ridge. The velocity–depth relation may cause an alteration front [Escartin et al., 2003; Python found by Canales et al. [2008] and the sonic log data from et al., 2007]. However, alteration would cause a gradual U1309D show strong similarities to our results for the depth change in velocity and density instead of a sharp contrast, as interval of 1–1.5 km. Velocities and the velocity gradient are derived from the joint forward modeling of mantle refrac- slightly higher, though (Figure 8). tions and gravity field data. Furthermore, the occurrence of [27] However, sampling on the southern wall of the ′ PmP reflections supports a significant velocity contrast at 22°19 N core complex sampled mainly serpentinized peri- ‐ dotites at the lower footwall (dive21), suggesting either that depth. The presence of gabbros, intensely sheared in high temperature conditions (amphibolite facies), sampled with the core complex is laterally heterogeneous or that the sur- Nautile on the steep southeastern wall of the dome (see face lithology did not represent the internal structure of discussion in section 3), is typical for magmatic processes. OCCs. Similar features have been found elsewhere. For Additionally, the sampled basalts (Figure 2) on top of the example, dredges from a rifted inside corner high at the core complex, in the median valley, and on the conjugated MAR near 5°S found that serpentinites have been smeared ridge flank indicate that magmatism took place while the as fault gouges along the corrugated slip surface but that the detachment fault was exposed. These features support the core complex internally was largely plutonic [Reston et al., interpretation that the active fault was rooted in a zone of 2002, Planert et al., 2010]. However, a micro‐earthquake magmatic accretion. The sonic log data of hole 735 show survey in that region [Tilmann et al., 2004] showed that a sharp transition exists between the magmatic segment center high velocities typical for gabbro. These data and the velocity–depth profile of the seismic velocity model from and the amagmatic segment end. This transition may reflect Muller et al. [2000] have been plotted in Figure 8. In con- a change in rheology and lithology. Thus, it may suggest that near ridge crest discontinuities some of the melt ex- trast to our OCC (dredged basalts and diabase), at the Atlantis Bank predominantly gabbroic rocks have been tracted from the asthenosphere crystallizes in the mantle and sampled. Thus, for comparison, the log data have been hence part of the crust is made of fractured and serpentinized

11 of 15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106 shifted downward below 1.5 km to compare the lithological rafted blocks (Figure 1c). Rafted blocks on top of the OCCs similar sections between both core complexes. For the have been described between 13° and 15°N [Smith et al., deeper part (deeper than 1.5 km), the velocity structure is 2008] and in the Canary Basin, Central Atlantic [Reston very similar to our final velocity model, and the velocities et al., 2004]. The occurrence of rafted blocks is supported are slightly slower. by diabase dykes and basalts sampled from the top, the [29] PmP reflections, observed in our seismic data northern and the eastern slopes of the massif (Figure 2). (Figures 3 and 4), and the density contrast obtained in our Along with our final seismic velocity and density model the modeling support crust that is roughly ∼40% thinner under observed features could generally be explained by a “rolling‐ the core complex (Figure 7), suggesting that accretion of the hinge” model proposed by Buck [1988]. The model assumes American plate is dominated by faulting rather than mag- that faults have ranges of optimal orientations to the principal matic construction during the past ∼2 million years. In stresses. According to the Mohr‐Coulomb criterion, shear contrast, crust under the median valley and eastern ridge failure will occur once the shear strength and internal fric- flank, representing the hanging wall of the fault system, is tion of a rock body are overcome. Rock mechanics experi- roughly 5 km thick, and seafloor topography shows gener- ments and numerical simulations show that when a fault is ally ridge parallel abyssal hill fabric. It is therefore reason- rotated away from the optimum angle for shear, frictional able to classify these areas as oceanic crust accreted sliding along a fault may become more difficult than creat- dominantly by magmatic processes. ing a new fault through unfractured rock [e.g., Nur et al., [30] Reduced seismic velocities in the upper mantle might 1986; Buck, 1988]. In this scenario, a normal fault, when be caused to some extent by high mantle temperatures and significantly rotated from the optimum angle of slip, relative the existence of a few percent of melt. Grevemeyer et al. to the crustal stress field, is replaced by a new planar fault [1998], for example, detected at the East Pacific Rise that orientated in the optimum direction. Thus, the low angle mantle velocities increase from <7.5 km/s near the ridge detachment fault remains inactive and is covered with rafted crest to >8 km/s in rough 10 m.y. old crust. At slow‐ blocks. The geometry of the detachment fault and uplift of spreading ridges, faults penetrating down to mantle depth the dome‐shaped core complex is further affected by the low [Huang and Solomon, 1988] may provide paths for seawater flexural rigidity of zero‐aged crust. to reach the upper mantle. Evidence from ophiolites sup- [33] Core complex formation at the Atlantis Bank at the ports the penetration of high‐temperature hydrothermal Southwest Indian Ridge caused a strong asymmetry in both fluids down to mantle depths [Leblanc et al., 1991; Benoit lithospheric structure and spreading rate. Baines et al. [2008] et al., 1999; Python et al., 2007]. Hydration and hence reported that 80% of plate motion was accommodated by the serpentinization may therefore affect seismic velocities. detachment fault, causing a northward migration of the ridge However, as mantle velocities seem to increase from ∼7.5 km/s crest with respect to adjacent segments spreading symmet- at the ridge crest to 7.8 km/s roughly 60 km off‐axis, we rically. Therefore, Baines et al. [2008] hypothesized that favor the interpretation that changes in the thermal regime asymmetric spreading rates may be a characteristic feature dominate this evolutionary process. of other detachment faults. Unfortunately, magnetic data from the 22°19′N core complex are not as detailed as from 6.3. Core Complex Formation the Atlantis Bank. However, the existing data from Gente [31] The critical question is whether the proposed detach- et al. [1995] suggest that lithospheric accretion at the ment fault causing the core complex is still active or inactive. MAR near 22°19′N occurs symmetrical with respect to the Unfortunately, no fault plane solutions are available for this spreading rate of the African and North American plates. area. However, existing fault plain solutions for different [34] Based on our seismic velocity model, bathymetry and areas at mid‐ocean ridges support that faulting generally seafloor ages from magnetic data [Gente et al., 1995], we occurs at normal faults dipping at 40°–60° [Huang and derived an evolutionary model for the core complex for- Solomon, 1987, 1988]. The distribution of earthquakes mation at 22°19′N (Figure 9). Roughly 1.9 m.y. ago magma from hydroacoustic monitoring indicate that near 22°19′N supply decreases and core complex formation initiated. In seismicity is concentrated below the axial valley [Smith our scenario, the initial fault dips at ∼60° and is located in et al., 2003], while the core complex itself is almost the median valley. When rotated away from its optimum aseismic. A similar feature has been observed in the TAG dip, slip stops, and a new fault is created. Tectonic extension area at 26°10′N using a local network of ocean bottom causes crustal thinning of 40% and exposed gabbroic lower seismic stations [de Martin et al., 2007]. In addition to the crust of about 20 km. Crustal thinning, flexural fault rota- distribution of seismicity within the area, the deployment tion, and perhaps serpentinization cause an uplift of the core provided earthquake hypocenters, indicating an active steep complex, resulting into the prominent dome‐shaped massif fault (∼70°) that is rooted in the median valley. Seismic of the 22°19′N core complex. velocities derived from crustal phases provided a velocity structure similar to our tomographic inversion, suggesting that lower crustal rocks are being exhumed in a detachment 7. Conclusions fault, which appears to roll over to a shallow dip of 20° and [35] Across the Mid‐Atlantic Ridge at 22°19′N seismic become aseismic at a depth of ∼3 km. This scenario is further refraction data show large lateral variations of the upper supported by seismic data from fossil oceanic detachment crustal P wave velocity structure. We observe the lowest faults [Ranero and Reston, 1999]. velocities in the median valley, increasing gradually toward [32] In contrast to the TAG detachment fault, bathymetric the east and rapidly near the western ridge flank. The data from the 22°19′N core complex suggest that the inac- western flank shows a prominent high of an oceanic core tive portion of the detachment fault surface is covered by complex. Velocities up to 4.5 km/s may indicate exposure of

12 of 15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106 highly fractured lower crustal rocks. Roughly 1.5 km below CL,F)2.0Si7.0All.0). Plagioclase in the matrix has the same the seabed, seismic velocities reach 7 km/s. Fast material An content as the one of porphyroclasts. These data indicate under the dome‐shaped core complex clearly support gab- that deformation occurred in conditions of high temperature broic rocks occurring at shallow depth, suggesting that the (typically around 500°C–600°C) and of water saturation. No rapid change in seismic velocities is related to a lithological evidence for overprint by a lower T° deformation episode is change caused by the unroofing of the crust. In contrast, found in this sample (no retromorphic mineral assemblages, seismic velocity data suggest that the lithosphere at the nor catactlastic deformation). The amphibolitized gabbro conjugated eastern flank was accreted by processes domi- can be interpreted as a witness of a ductile, possibly detach- nated by magmatic construction and hence show a typical ment fault, rooted in a former magma body that accommo- velocity depth distribution expected for normal oceanic dated shearing at high temperature. crust. [40] Although 100% serpentinized (which prevented any [36] Joint raytracing and gravity modeling suggest that geochemical analysis of the primary minerals), the serpen- crust under the core complex is approximately 2 km thinner tinized peridotites have preserved a coarse‐grained (up to than under the conjugated ridge flank. The velocity and 1 cm) mosaic textures diagnostic of hot working in low density contrast as well as a few clear PmP reflections stress and high temperature (∼1200°C) conditions. Such support that a Moho boundary was created during core textures are classically attributed to deformation related to complex formation, suggesting that the detachment fault is “asthenospheric” mantle flow [Ceuleneer et al., 1988]. The rooted in a zone of magmatic accretion at 2–4 km depth. orientation of the deformation plane, measured on one [37] We favor the interpretation that the core complex was sample reoriented on board is subhorizontal, a situation formed by a mechanism similar to the rolling‐hinge model similar to the one observed in the nearby peridotite massif by Buck [1988]. The core complex formation started drilled at DSDP Site 920 [Ceuleneer and Cannat, 1997]. No roughly 1.9 m.y. ago and is perhaps still active. Even though overprint of lower temperature deformation (500°C–600°C), accretion seems to be highly asymmetric in terms of the equivalent to the one recorded by the foothill amphibolitized lithospheric structure of conjugated ridge flanks, accretion gabbro, has been observed on this outcrop. The rock is of North American and African plates occur at similar rates. harzburgitic in composition (i.e., a prealteration assemblage A major discontinuity could explain the sharp transition of olivine and orthopyroxene). Interstitial grains (likely between the crustal lithology beneath the massif and ultra- former clinopyroxene) diagnostic of precipitation from a mafics occurring at the southern wall [Karson et al., 2006]. melt are observed in a few samples. However, this transition might be controlled by the thermal structure, favoring melt being trapped in the mantle during [41] Acknowledgments. We are grateful for the support of captain Kull and his crew of RV Meteor for their excellent support during the expe- melt extraction at cool segment ends [e.g., Cannatetal., dition M60/2. The manuscript benefited from the review of two anonymous 1995; Cannat, 1996]. referees. The research was funded by the German Science Foundation (DFG) (grants Mo 961‐5/1 Ra 925‐5/1 Gr 1964‐8/2+8/3). Appendix

[38] The mylonitic gabbro sampled at the base of the References section contains small porphyroclasts of clinopyroxene Baines, A. G., M. J. Cheadle, B. E. John, and J. J. Schwartz (2008), The (cpx) and of plagioclase (plg) up to 0.5 mm in size; they are rate of oceanic detachment faulting at Atlantis Bank, SW Indian Ridge, – embedded in a fine‐grained matrix made of a mosaic of plg Earth Planet. Sci. Lett., 273, 105 114, doi:10.1016/j.epsl.2008.06.013. Benoit, M., G. Ceuleneer, and M. Polvé (1999), The remelting of hydro- and amphibole a few tens of microns in size with sutured thermally altered peridotite at mid‐ocean ridges by intruding mantle diapir, grain boundaries. Plg porphyroclasts present well‐developed Nature, 402, 514–518, doi:10.1038/990073. mechanical twins. Cpx porphyroclasts have generally Birch, F. (1961), The velocity of compressional waves in rocks to 10 kilobars, part 2, J. Geophys. Res., 66(B7), 2199–2224, doi:10.1029/ rounded shapes and are heavily recrystallized and trans- JZ066i007p02199. formed into hornblende on their border. The protolith of this Blackman, D. K., J. R. Cann, B. Janssen, and D. K. Smith, (1998), Origin mylonite was a gabbro devoid of olivine and of oxides. The of extensional core complexes: Evidence from the Mid‐Atlantic Ridge at Atlantis fracture zone, J. Geophys. Res., 103(B9), 21,315–21,333. porphyroclasts allow us to infer the primary igneous phase Blackman, D. K., B. Ildefonse, B. E. John, Y. Ohara, D. J. Miller, and C. J. composition. Plg porphyroclasts have an average An content MacLeod (2006), Proc. Integrated Ocean Drilling Program, vols. 304/ of 0.55. Some porphyroclasts present a slight inverse zoning 305, The Geological Society of America, College Station, Texas. with An ranging from 0.48 to 0.57, but most of them are Buck, W. R. (1988), Flexural rotation of normal faults, Tectonics, 7(5), 959–973, doi:10.1029/TC007i005p00959. quite homogeneous in composition. Cpx porphyroclasts are Canales, J. P., and J. A. Collins, (2000), Seismic structure across the rift augites with a Mg# of 0.78, a low Cr2O3 content (0.1 wt%), valley of the Mid‐Atlantic Ridge at 23°20′ (MARK area): Implications a moderate Al O content (2.5 wt%), and a high TiO for crustal accretion processes at slow spreading ridges, J. Geophys. 2 3 2 Res., 105(B12), 28,411–28,425. content (0.7 wt%). These values are typical of evolved cu- Canales, J. P., B. E. Tucholke, and J. A. Collins (2004), Seismic reflection mulates from normal mid‐oceanic ridge basalt (NMORB) imaging of an oceanic detachment fault: Atlantis megamullion (Mid‐ [e.g., Elthon, 1987; Natland and Dick, 1996]. With respect Atlantic Ridge, 30°10′N), Earth Planet. Sci. Lett., 222,543–560, to gabbros drilled in the nearby ODP Site 923, the samples doi:10.1016/j.epsl.2004.02.023. Canales, J. P., B. E. Tucholke, M. Xu, J. A. Collins, and D. L. DuBois follow the differentiation trend, indicating that they derive (2008), Seismic evidence for large‐scale compositional heterogeneity from similar parent melts. The composition of these gabbros of oceanic core complexes, Geochem. Geophys. Geosyst., 9, Q08002, is the one expected for cumulates from mid‐oceanic ridge doi:10.1029/2008GC002009. Cann, J. R., D. K. Blackman, D. K. Smith, E. McAllister, B. Janssen, basalt (MORB) of the MARK area [Ross and Elthon, 1997]. S. Mello, E. Avgerinos, A. R. Pascoe, and J. Escartin (1997), Corrugated [39] Amphibole from the matrix is a common green slip surfaces formed at ridge‐transform intersections on the Mid‐Atlantic Ridge, Nature, 385, 329–332, doi:10.1038/385329a0. hornblende (average formula is Na0.5Ca1.8(Mg,Fe)4.7(OH,

13 of 15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Cannat, M. (1996), How thick is the magmatic crust at slow spreading Karson, J. A., et al. (1987), Along‐axis variations in seafloor spreading in ridges? J. Geophys. Res., 101(B2), 2847–2857, doi:10.1029/95JB03116. the MARK area, Nature, 328, 681–685, doi:10.1038/328681a0. Cannat, M., et al. (1995), Thin crust, ultramafic exposures, and rugged Karson, J. A., G. L. Früh‐Green, D. S. Kelley, E. A. Williams, D. R. Yoerger, faulting patterns at the Mid‐Atlantic Ridge (22°–24°N), Geology, 23, and M. Jakuba (2006), Detachment shear zone of the Atlantis Massif core 49–52, doi:10.1130/0091-7613(1995)023<0049:TCUEAR>2.3.CO;2. complex, Mid‐Atlantic Ridge, 30°N, Geochem. Geophys. Geosyst., 7, Carlson, R. L. (1998), Seismic velocities in the uppermost oceanic crust: Q06016, doi:10.1029/2005GC001109. Age dependence and the fate of layer 2A, J. Geophys. Res., 103(B4), Korenaga, et al. (2000), Crustal structure of the southeast Greenland margin 7069–7077, doi:10.1029/97JB03577. from joint refraction and reflection seismic tomography, J. Geophys. Carlson, R. L., and C. N. Herrick (1990), Densities and porosities in the Res., 105(B9), 21,591–21,614, doi:10.1029/2000JB900188. oceanic crust and their variations with depth and age, J. Geophys. Res., Leblanc, M., G. Ceuleneer, H. Al Azri, and J. Jedwab (1991), Concentra- 95(B6), 9153–9170, doi:10.1029/JB095iB06p09153. tion hydrothermale de Pd et de Pt dans les péridotites mantellaires du Carlson, R. L., and D. J. Miller (2003), Mantle wedge water contents esti- complexe ophiolitique d’Oman, C.R. Acad. Sci., 312(II), 1007–1012. mated from seismic velocities in partially serpentinised peridotites, Geo- Luetgert, J. H. (1992) Interactive two‐dimensional seismic raytracing for phys. Res. Lett., 30(5), 1250, doi:10.1029/2002GL016600. the Macintosh, U.S. Geological Survey, Open‐File Report 92–356, 44. Ceuleneer, G., and M. Cannat (1997), High temperature ductile deformation Maia, M., and P. Gente (1998), Three‐dimensional gravity and bathymetry of the site 920 peridotites, in Proc. Ocean Drilling Program, Scientific analysis of the Mid‐Atlantic Ridge between 20°N and 24°N: Flow Results, vol. 153, edited by J. A. Karson, M. Cannat, D. J. Miller, and geometry and temporal evolution of the segmentation, J. Geophys. D. Elthon, pp. 23–34, Ocean Drilling Program, College Station, Tex., Res., 103(B1), 951–974, doi:10.1029/97JB01635. doi:10.2973/odp.proc.sr.153.002.1997. Muller, M. R., T. A. Minshull, and R. S. White (2000), Crustal structure of Ceuleneer, G., A. Nicolas, and F. Boudier (1988), Mantle flow patterns at the Southwest Indian Ridge at the Atlantis II fracture zone, J. Geophys. an oceanic spreading centre: The Oman peridotites record, Tectonophysics, Res., 105(B11), 25,809–25,828, doi:10.1029/2000JB900262. 151,1–26, doi:10.1016/0040-1951(88)90238-7. Natland, J. H., and H. J. B. Dick (1996), Melt migration through high‐level de Martin, B. J., R. A. Sohn, J. P. Canales, and S. E. Humphris (2007), gabbroic cumulates of the East Pacific rise at Hess deep: The origin of Kinematics and geometry of active detachment faulting beneath the magma lenses and the deep crustal structure of fast spreading ridges, in Trans‐Atlantic Geotraverse (TAG) hydrothermal field on the Mid‐Atlantic Proc. Ocean Drilling Program, Scientific Results vol. 147, edited by Ridge, Geology, 35, 711–714, doi:10.1130/G23718A.1. C. Mevel, K. M. Gillis, J. F. Allan, and D. Elthon, pp. 21–59, College Dick, H. J. B., et al. (2000), A long in situ section of the lower ocean crust: Station, Ocean Drilling Program, Texas, doi:10.2973/odp.proc.sr.147. Results of ODP Leg 176 drilling at the Southwest Indian Ridge, Earth 002.1996. Planet. Sci. Lett., 179,31–51, doi:10.1016/S0012-821X(00)00102-3. Nedimović, M. R., S. M. Carbotte, J. B. Diebold, A. J. Harding, J. P. Canales, Dick, H. J. B., M. A. Tivey, and B. E. Tucholke (2008), Plutonic founda- and G. M. Kent (2008), Upper crustal evolution across the Juan de Fuca tion of a slow‐spreading ridge segment: Oceanic core complex at Kane ridge flanks, Geochem. Geophys. Geosyst., 9, Q09006, doi:10.1029/ Megamullion, 23°30′N, 45°20′W, Geochem. Geophys. Geosyst., 9, 2008GC002085. Q05014, doi:10.1029/2007GC001645. Nur, A., H. Ron, and O. Scotti (1986), Fault mechanics and the kinematics Elthon, D. (1987), Petrology of gabbroic rocks from the Mid‐Cayman Rise of block rotations, Geology, 14, 746–749, doi:10.1130/0091-7613(1986) spreading center, J. Geophys. Res., 92(B1), 658–682, doi:10.1029/ 14<746:FMATKO>2.0.CO;2. JB092iB01p00658. Planert, L., E. R. Flueh, F. Tilmann, I. Grevemeyer, and T. J. Reston Escartin, J., C. Mével, C. J. MacLeod, and A. M. McCaig (2003), Con- (2010), Crustal structure of a rifted oceanic core complex and its conju- straints on deformation conditions and the origin of oceanic detachments: gate side at the MAR at 5°S: Implications for melt extraction during The Mid‐Atlantic Ridge core complex at 15°45′N, Geochem. Geophys. detachment faulting and core complex formation, Geophys. J. Int., 181, Geosyst., 4(8), 1067, doi:10.1029/2002GC000472. 113–126. Escartin, J., D. K. Smith, J. Cann, H. Shouten, C. L. Langmuir, and S. Escrig Purdy, G. M., and R. S. Detrick (1986), Crustal structure of the Mid‐Atlantic (2008), Central role of detachment faults in accretion of slow spreading Ridge at 23°N from seismic refraction studies, J. Geophys. Res., 91(B3), oceanic lithosphere, Nature, 455, 790–794, doi:10.1038/nature07333. 3739–3762, doi:10.1029/JB091iB03p03739. Flueh, E. R., and J. Bialas (1996), A digital, high data capacity ocean bottom Python, M., G. Ceuleneer, Y. Ishida, J. A. Barrat, and S. Arai (2007), Oman recorder for seismic investigations, Int. Underwater Syst. Design, 18, diopsidites a new lithology diagnostic of very high temperature hydro- 18–20. thermal circulation in mantle peridotite below oceanic spreading centres, Gente, P., et al. (1991), Geometry of past and present‐day segmentation of Earth Planet. Sci. Lett., 255, 289–305, doi:10.1016/j.epsl.2006.12.030. the Mid‐Atlantic Ridge south of Kane fracture zone, Eos Trans, AGU, Ranero, C. R., and T. J. Reston (1999), Detachment faulting at ocean core 72(17), Spring Meet. Suppl., 477. complexes, Geology, 27,983–986, doi:10.1130/0091-7613(1999) Gente, P., R. A. Pockalny, C. Durand, C. Deplus, M. Maia, G. Ceuleneer, 027<0983:DFAOCC>2.3.CO;2. C. Mével, M. Cannat, and C. Laverne (1995), Characteristics and evolu- Reston, T. J., W. Weinrebe, I. Grevemeyer, E. R. Flueh, N. C. Mitchell, tion of the segmentation of the Mid‐Atlantic Ridge between 20°N and L. Kirstein, C. Kopp, H. Kopp, et al. (2002), A rifted inside corner 24°N during the last 10 million years, Earth Planet. Sci. Lett., 129, massif on the Mid‐Atlantic Ridge at 5°S, Earth Planet. Sci. Lett., 200, 55–71, doi:10.1016/0012-821X(94)00233-O. 155–169, doi:10.1016/S0012-821X(02)00636-2. Ghose, I., M. Cannat, and M. Seyler (1996), Transform fault effect on mantle Reston, T. J., C. R. Ranero, O. Ruoff, M. Peres‐Gussinye, and J. J. Dañobeitia melting in the MARK area (Mid‐Atlantic Ridge south of the Kane trans- (2004), Geometry of extensional faults developed at slow‐spreading cen- form), Geology, 24,1139–1142, doi:10.1130/0091-7613(1996) tres from pre‐stacked depth migration of seismic reflection data in the 024<1139:TFEOMM>2.3.CO;2. Central Atlantic (Canary Basin), Geophys. J. Int., 159,591–606, Grevemeyer, I., and W. Weigel (1996), Seismic velocities of the uppermost doi:10.1111/j.1365-246X.2004.02444.x. igneous crust versus age, Geophys. J. Int., 124,631–635, doi:10.1111/ Ross, K., and D. Elthon (1997), Cumulus and postcumulus crystallization j.1365-246X.1996.tb07041.x. in the oceanic crust: Major and trace‐element geochemistry of Leg Grevemeyer, I., W. Weigel, and C. Jennrich (1998), Structure and ageing of 153 gabbroic rocks, in Proc. Ocean Drilling Program, Scientific Results, oceanic crust at 14°S on the East Pacific Rise, Geophys. J. Int., 135, 573– vol. 153, edited by J. A. Karson, M. Cannat, D. J. Miller, and D. Elthon, 584, doi:10.1046/j.1365-246X.1998.00673.x. pp. 23–34, College Station, Ocean Drilling Program, Texas, doi:10.2973/ Grevemeyer, I., N. Kaul, H. Villinger, and W. Weigel (1999), Hydrother- odp.proc.sr.153.023.1997. mal activity and the evolution of the seismic properties of upper oceanic Smith, D. K., J. Escartin, M. Cannat, M. Tolstoy, C. G. Fox, D. R. Bohnenstiel, crust, J. Geophys. Res., 104(B3), 5069–5079, doi:10.1029/ and S. Basin (2003), Spatial and temporal distribution of seismicity along 1998JB900096. the Mid‐Atlantic Ridge (15°–35°N), J. Geophys. Res., 108(B3), 2167, Huang, P. Y., and S. C. Solomon (1987), Centroid depths and mechanisms doi:10.1029/2002JB001964. of Mid‐Ocean Ridge earthquakes in the Indian Ocean, Gulf of Aden, and Smith, D. K., J. Escartın, H. Schouten, and J. R. Cann (2008), Fault rotation Red Sea, J. Geophys. Res., 92(B2), 1361–1382, doi:10.1029/ and core complex formation: Significant processes in seafloor formation JB092iB02p01361. at slow‐spreading mid‐ocean ridges (Mid‐Atlantic Ridge, 13°–15°N), Huang, P. Y., and S. C. Solomon (1988), Centroid depths of mid‐ocean Geochem. Geophys. Geosyst., 9, Q03003, doi:10.1029/2007GC001699. ridge earthquakes: Dependence on spreading rate, J. Geophys. Res., Smith, D. K., J. R. Cann, and J. Escartin (2006), Widespread active detach- 93(B11), 13,445–13,477, doi:10.1029/JB093iB11p13445. ment faulting and core complex formation near 13°N on the Mid‐Atlantic Ildefonse, B., D. K. Blackman, B. E. John, Y. Ohara, D. J. Miller, and C. J. Ridge, Nature, 442, 440–443, doi:10.1038/nature04950. MacLeod (2007), Oceanic core complexes and crustal accretion at slow‐ Stevenson, J. M., J. A. Hildebrand, and M. A. Zumberge (1994), An ocean spreading ridges, Geology, 35, 623–626, doi:10.1130/G23531A.1. bottom gravity study of the southern Juan de Fuca Ridge, J. Geophys. Res., 99(B3), 4875–4888, doi:10.1029/93JB02076.

14 of 15 B07106 DANNOWSKI ET AL.: OCC AT MAR—SEISMIC STUDY B07106

Tilmann,F.,E.Flueh,L.Planert,T.Reston,andW.Weinrebe(2004), White, R. S., D. McKenzie, and R. K. O’Nions, (1992), Oceanic crustal Microearthquake seismicity of the Mid‐Atlantic Ridge at 5°S: A view thickness from seismic measurements and rare earth element inversions, of tectonic extension, J. Geophys. Res., 109(B6), B06102, doi:10.1029/ J. Geophys. Res., 97(B13), 19 683–19 715. 2003JB002827. Toomey, D. R., S. C. Solomin, and G. M. Purdy (1988), Microearthquakes A. Dannowski and I. Grevemeyer, Leibniz Institut für beneath the median valley of the Mid‐Atlantic Ridge near 23°N: Tomog- ‐ ‐ – Meereswissenschaften, IFM GEOMAR, Wischhofstr. 1 3, 24 148 Kiel, raphy and tectonics, J. Geophys. Res., 93(B8), 9093 9112, doi:10.1029/ Germany. (adannowski@ifm‐geomar.de) JB093iB08p09093. C. R. Ranero, ICREA at Institut de Ciencias del Mar, CSIC, Pg. Marítim Tucholke, B. E., and J. Lin (1994), A geological model for the structure of de la Barceloneta 37‐49, 08003 Barcelona, Spain. ridge segments in slow spreading ocean crust, J. Geophys. Res., 99(B6), ‐ – G. Ceuleneer, Toulouse University, Observatoire Midi Pyrénées, 11,937 11,958, doi:10.1029/94JB00338. CNRS‐UMR5562, 14 Avenue E. Belin, 31400 Toulouse, France. Tucholke, B. E., J. Lin, and M. C. Kleinrock (1998), Megamullions and M. Maia and P. Gente, Université Européenne de Bretagne, CNRS mullion structure defining oceanic metamorphic core complexes on the ‐ – UMR 6538, Brest, France. Mid Atlantic Ridge, J. Geophys. Res., 103(B5), 9857 9866, J. P. Morgan, EAS Department, Cornell University, Ithaca, NY doi:10.1029/98JB00167. 14853, USA.

15 of 15